Predicting flow-generated noise from HVAC components

University dissertation from Stockholm : KTH Royal Institute of Technology

Abstract: More energy efficient fans, i.e. larger sizes running at lower speeds, in Heating Ventilation and Air Conditioning (HVAC) systems decrease the fan noise and increase the importance of flow generated noise in other system components, e.g., dampers and air terminal devices. In this thesis, an extended prediction model, using semi-empirical scaling laws, for flow noise prediction in HVAC systems at low Mach number flow speeds is presented. The scaling laws can be seen as a combination of a generalized noise spectrum based on experimental data and constriction flow characteristics, where the latter can be gained from ComputationalFluid Dynamics (CFD) simulations. The flow generated noise can be predicted by semi-empirical scaling laws to avoid a time consuming, fully resolved simulation or measurement. Here, an approach is suggested where the general noise spectra are combined with turbulent data obtained from Reynolds Average Navier Stokes (RANS) simulations. A model is proposed using a momentumflux assumption of the dipole source strength and a frequency scaling based on the constriction pressure loss. To evaluate the applicability of the semi-emprical scaling law on different HVAC geometries both literature data and new measurement data are considered. Focus is at comparing geometries of high and low pressure loss but also to discuss the differences in other properties, e.g. radiation characteristics. A general noise reference spectrum is determined bya best fit calculation of measurement data including orifice, damper and bend geometries. Air terminal devices at the end of a duct are also evaluated and compared to constrictions inside ducts. The expected accuracy of the suggested model and its challenges as a tool for flow noise prediction of non-rotating components in HVAC systems are discussed.